Dataset and code associated to "Geometric inflation of deviations challenges neural quantum states in dynamics of quantum Ising models"
Authors/Creators
Description
This Zenodo entry contains the data, scripts and source code needed to
reproduce the numerical experiments and figures of the publication
**"Geometric inflation of deviations challenges neural quantum states in
dynamics of quantum Ising models"** (W. Krinitsin, J. B. Rigo, M. Abedi,
M. Schmitt):
https://doi.org/10.48550/arXiv.2609.07645
The folder layout follows the figures of the paper. Each numerical
experiment lives in its own subdirectory and follows the layout produced by
encap (https://github.com/computational-quantum-science/encap), a thin
wrapper that copies the main script and its auxiliary files into a
self-contained run directory (all experiments except the supervised-learning
runs of Figures 2–4, which are plain Python scripts driven by a JSON
configuration):
* `run.py` (or `run.jl`) — the main script
* `input.json` — all parameters of the run
* `mytdvp.py`, `nets/`, `sampler/` — auxiliary Python modules used by `run.py`
* `*.csv`, `*.h5`, `*.npy` — data produced by the run
Experiment folders are named `<encap script>__<encap version>`:
`crit_tfim_update__*` are TDVP time evolutions / evaluations with jVMC,
`crit_tfim_fidelities__*` compute fidelities with the exact state, and
`exact_krylov__*` are the exact (quspin, Lanczos) reference evolutions.
The plots themselves are generated by the Julia notebook `plots.ipynb` at
the top level of this entry. All paths inside the notebook and inside each
script are relative to this folder; nothing outside it is needed apart from
the libraries listed in `README.md` (pinned versions in
`requirements_jvmc.txt` and `requirements_quspin.txt`).
Top-level layout:
```
zenodo/
├── README.md
├── plots.ipynb Julia notebook that produces all figures
├── requirements_jvmc.txt Pinned Python packages (jVMC / JAX environment)
├── requirements_quspin.txt Pinned Python packages (quspin environment)
├── figures/ Output PDFs of plots.ipynb
├── schematics/ Scripts + SVGs of the hand-drawn panels
├── figure1_nqs_failure/ Figure 1 d,e
├── figure2_3_4_supervised_learning/ Figures 2, 3, 4 (supervised learning, one experiment)
├── figure5_backward_forward_evolution/ Figure 5
├── figure6_sensitivity/ Figure 6 b–e
├── figure7_nullspace_rotation/ Figure 7 a,b (and data of Fig. 12, 13)
├── figure8_tdvp_hyperparameters/ Figure 8
├── figure9_time_reversal_regularization/ Figure 9
├── figure10_backward_forward_appendix/ Figure 10
├── figure11_mps_infidelity/ Figure 11
├── figure12_s_matrix_spectra/ Figure 12 (README only)
└── figure13_nullspace_rotation_appendix/ Figure 13 (README only)
```
Each subfolder contains its own `README.md` describing the experiments and
how they feed into the corresponding figure.